Study Of Image Processing Based On Saliency Detection For Artificial Vision

Humans perceive the world through the eyes.Blindness will not only bring a lot of trouble to the life of patients,but also bring pressure to the family life.In order to alleviate the troubles of blind people's life,and the pain of family and the pressure of society,it is imperative to partially restore the sight of blind people.Currently,neuropathic retinal diseases such as primary age-related macular degeneration and retinitis pigmentosa still have no effective drug therapy or treatment in surgery.For these kind of eye diseases,retinal prosthesis replaces the impaired part and stimulates the adjacent intact tissue of the visual pathway,so that the functional vision of the blind can be restored to a certain extent.However,the number of implantable electrodes is very limited at present,which leads to a great loss of external information provided by the prosthesis.Therefore,an effective image processing algorithm is introduced into the visual information processing module of the prosthesis device to optimize the information expressed by the electrode array at finite resolution,which is a feasible method to solve the above problems.At present,image processing algorithms used to improve the performance of retinal prosthesis devices have been widely studied under the simulating condition of artificial vision.However,in order to get more information of the region of interest,most algorithms cannot realize real-time processing due to the complex computation,which limits the possibility of their practical application.In order to meet the needs of implant users in daily life,real-time processing and optimal expression of image information is necessary.In view of the above problems,this study proposed a real-time image processing algorithm for information extraction of the interested region under artificial vision,by utilizing the luminance contrast as the basic feature which the human visual perception system is very sensitive to.Combined with the color space conversion processing,the saliency computation based on the global luminance contrast of the image was used to obtain the salient object detection results in the captured image.In order to obtain more accurate salient object information,the paper proposed a "visual attention simulation processing" model.Based on the theory of the visual features of human eyes,the model adaptively extracted the most salient regions using a gaussian filter.In this way,the background noise caused by foreground enhancement was filtered out,and the foreground information missing result from only use the global luminance contrast was completed,and the salient information in the image was effectively extracted.Experiments were carried out to evaluate the proposed method based on global luminance contrast.Firstly,the algorithm in the paper was evaluated in a quantitative and qualitative way through two public benchmark image databases.The evaluation results verified the superiority of the algorithm in the aspect of extracting image salient information compared with other algorithms,and the processing time of the algorithm was also tested.On the other hand,the study implemented the improved saliency computation method on the artificial vision simulation experimental platform of retinal prosthesis,selected pictures that something the prosthesis implanter used in daily life as experimental materials.Meantime,participants were recruited to carry out the simulated artificial vision experiments.The experimental results,compared with the original images and images processed by the basic global luminance contrast image processing,showed that the improved method proposed in the paper could optimize the information expression,and improved the object recognition efficiency under artificial vision in the simulation environment.The research work would be a reference for the further application of the proposed method in the retinal prosthesis,and would provide an important experimental basis for the postoperative training for visual function rehabilitation of retinal prosthesis implant patients.